Project Summary: RNA serves important roles as an information carrier and effector molecule. For many roles, a functional RNA must adopt a specific secondary or tertiary structure. Across evolution, these structures are more conserved than the sequence. By comparing multiple homologous sequences, structures can be inferred. We developed TurboFold as an accurate and rapid method for automating sequence comparison to predict conserved RNA secondary structures and alignments. We also expanded this with a knowledge-based potential to predict conserved non-canonical base pairs, which are the basis of tertiary structures. Expanding on our TurboFold method, we will develop new high-impact algorithms and software to solve important problems in RNA biology. First, we will develop new tools to estimate phylogenies for RNA and to use the phylogenetic relationships between sequences to more accurately predict structures and alignments. Second, we will improve tertiary structure modeling by using our predictions of conserved non-canonical pairs as restraints for building all-atom models. Third, we will develop new tools for homology modeling of secondary and tertiary structure, where a template structure for an RNA sequence from the same family exists.